bug: investigate what's (purportedly) hogging resources

[freelerobot]

Describe the bug
A clear and concise description of what the bug is.

A few reports of Jan v0.4.4 taking up compute and or ram.

Steps to reproduce
Steps to reproduce the behavior:

1. Download and compare 0.4.3 vs 0.4.3

Expected behavior
A clear and concise description of what you expected to happen.

1. Jan shouldn't actively use RAM & CPU when idle
2. It could be that users are confusing System Monitor metrics to mean that "Jan is using 80% of my CPU/RAM"
3. When, in fact, we're just measuring all RAM usage across all applications.

Regardless, let's treat this as a performance regression and investigate it as such.

Outcomes

1. Diagnostic on what is consuming RAM
2. Diagnostic on what is consuming CPU
3. Overview of Jans main processes and % of utilization (when actively being used)
4. A list of Jan processes that can improved (regardless if we are being a resource hog or not)

Screenshots
If applicable, add screenshots to help explain your issue.

Users may be confused and think this is Jan utilization, when in fact, it is all applications.
@imtuyethan 's new designs should partly address this:

Environment details

• Assume all envs

Logs
N/A

Additional context
Add any other context or information that could be helpful in diagnosing the problem.

[hiento09]

This file https://github.com/janhq/jan/blob/main/extensions/inference-nitro-extension/src/module.ts, line 103 cause this bug, @0xSage , @Van-QA . The issue is described as follows: in version 0.4.4, the Nitro CPU utilizes all of the number of CPU cores for inference, resulting in 100% CPU usage. This leads to a lack of CPU resources for the Jan app to process other tasks, consequently slowing down the user's entire machine. To fix this issue, I will create a Pull Request (PR) to modify the CPU version to use only half of the CPU cores for running Nitro, instead of using all cores as it currently does

[freelerobot]

Even when there’s no inference happening, ie when the app is idle? It makes sense that we use CPU when inference is happening. My concern is when the app is idle that it’s still hogging resources. I’m not comfortable with the hotfix. It is very hacky. Let’s discuss a long term solution for this.

…

On Thu, Jan 18, 2024 at 6:35 PM hiento09 ***@***.***> wrote: This file https://github.com/janhq/jan/blob/main/extensions/inference-nitro-extension/src/module.ts, line 103 cause this bug, @0xSage <https://github.com/0xSage> , @Van-QA <https://github.com/Van-QA> . The issue is described as follows: in version 0.4.4, the Nitro CPU utilizes all of the number of CPU cores for inference, resulting in 100% CPU usage. This leads to a lack of CPU resources for the Jan app to process other tasks, consequently slowing down the user's entire machine. To fix this issue, I will create a Pull Request (PR) to modify the CPU version to use only half of the CPU cores for running Nitro, instead of using all cores as it currently does image.png (view on web) <https://github.com/janhq/jan/assets/136591877/5b451386-ab57-483e-9a08-eca5ac409ebb> — Reply to this email directly, view it on GitHub <#1650 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQVWFCFKMWBGOVTI3A5FB2TYPD3HJAVCNFSM6AAAAABB63DE26VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQOJYGIYTQMRTGU> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[freelerobot]

@hiro-v you might wanna convert this into an epic to properly track all the follow ups in the Tasklist

[hiro-v]

Closing in favor of #1714